Dynamic Random Access Memory utilizes capacitors to store bits of information within an integrated circuit. A capacitor is formed by placing a dielectric material between two electrodes formed from conductive materials. A capacitor's ability to hold electrical charge (i.e., capacitance) is a function of the surface area of the capacitor plates A, the distance between the capacitor plates d (i.e. the physical thickness of the dielectric layer), and the relative dielectric constant or k-value of the dielectric material. The capacitance is given by:
                    C        =                  κ          ⁢                                          ⁢                      ɛ            o                    ⁢                      A            d                                              (                  Eqn          .                                          ⁢          1                )            where εo represents the vacuum permittivity.
The dielectric constant is a measure of a material's polarizability. Therefore, the higher the dielectric constant of a material, the more charge the capacitor can hold. Therefore, if the k-value of the dielectric is increased, the area of the capacitor can be decreased and maintain the desired cell capacitance. Reducing the size of capacitors within the device is important for the miniaturization of integrated circuits. This allows the packing of millions (mega-bit (Mb)) or billions (giga-bit (Gb)) of memory cells into a single semiconductor device. The goal is to maintain a large cell capacitance (generally ˜10 to 25 fF) and a low leakage current (generally <10−7 A cm−2). The physical thickness of the dielectric layers in DRAM capacitors could not be reduced unlimitedly in order to avoid leakage current caused by tunneling mechanisms which exponentially increases as the thickness of the dielectric layer decreases.
Traditionally, SiO2 has been used as the dielectric material and semiconducting materials (semiconductor-insulator-semiconductor [SIS] cell designs) have been used as the electrodes. The cell capacitance was maintained by increasing the area of the capacitor using very complex capacitor morphologies while also decreasing the thickness of the SiO2 dielectric layer. Increases of the leakage current above the desired specifications have demanded the development of new capacitor geometries, new electrode materials, and new dielectric materials. Cell designs have migrated to metal-insulator-semiconductor (MIS) and now to metal-insulator-metal (MIM) cell designs for higher performance.
One class of high-k dielectric materials possessing the characteristics required for implementation in advanced DRAM capacitors are high-k metal oxide materials. Examples of suitable dielectric materials comprise Al2O3, BaSrTiOx (BST), HfO2, HfSiOx, Nb2O5, PbZrTiOx (PZT), SiO2, a bilayer of SiO2 and SixNy, SiON, SrTiO3 (STO), Ta2O5, TiO2, ZrO2, etc. TiO2 and ZrO2 are two specific examples of metal oxide dielectric materials which display significant promise in terms of serving as a high-k dielectric material for implementation in DRAM capacitors.
Typically, DRAM devices at technology nodes of 80 nm and below use MIM capacitors wherein the electrode materials are metals. These electrode materials generally have higher conductivities than the semiconductor electrode materials, higher work functions, exhibit improved stability over the semiconductor electrode materials, and exhibit reduced depletion effects. The electrode materials must have high conductivity to ensure fast device speeds. Representative examples of electrode materials for MIM capacitors are metals, conductive metal oxides, conductive metal silicides, conductive metal nitrides (i.e. TiN), or combinations thereof. MIM capacitors in these DRAM applications utilize insulating materials having a dielectric constant, or k-value, significantly higher than that of SiO2 (k=3.9). For DRAM capacitors, the goal is to utilize dielectric materials with k values greater than about 40. Such materials are generally classified as high-k materials. Representative examples of high-k materials for MIM capacitors are non-conducting metal oxides, non-conducting metal nitrides, non-conducting metal silicates or combinations thereof. These dielectrics may also include additional dopant materials.
A figure of merit in DRAM technology is the electrical performance of the dielectric material as compared to SiO2 known as the Equivalent Oxide Thickness (EOT). A high-k material's EOT is calculated using a normalized measure of silicon dioxide (SiO2k=3.9) as a reference, given by:
                              E          ⁢                                          ⁢          O          ⁢                                          ⁢          T                =                              3.9            κ                    ·          d                                    (                  Eqn          .                                          ⁢          2                )            where d represents the physical thickness of the capacitor dielectric.
As DRAM technologies scale below the 40 nm technology node, manufacturers must reduce the EOT of the high-k dielectric films in MIM capacitors in order to increase charge storage capacity. The goal is to utilize dielectric materials that exhibit an EOT of less than about 0.8 nm while maintaining a physical thickness of about 5-20 nm.
Generally, as the dielectric constant of a material increases, the band gap of the material decreases. For example. The rutile phase of TiO2 has a k-value of about 80 and a band gap of about 3.0 eV while ZrO2 in the tetragonal phase has a k-value of about 43 and a band gap of about 5.8 eV. The low band gap may lead to high leakage current in the device. As a result, without the utilization of countervailing measures, capacitor stacks implementing high-k dielectric materials may experience large leakage currents. High work function electrodes (e.g., electrodes having a work function of greater than 5.0 eV) may be utilized in order to counter the effects of implementing a reduced band gap high-k dielectric layer within the DRAM capacitor. Metals, such as platinum, gold, ruthenium, and ruthenium oxide are examples of high work function electrode materials suitable for inhibiting device leakage in a DRAM capacitor having a high-k dielectric layer. The noble metal systems, however, are prohibitively expensive when employed in a mass production context. Moreover, electrodes fabricated from noble metals often suffer from poor manufacturing qualities, such as surface roughness, poor adhesion, and form a contamination risk in the fab.
Leakage current in capacitor dielectric materials can be due to Schottky emission, Frenkel-Poole defects (e.g. oxygen vacancies (Vox) or grain boundaries), or Fowler-Nordheim tunneling. Schottky emission, also called thermionic emission, is a common mechanism and is the thermally activated flow of charge over an energy barrier whereby the effective barrier height of a MIM capacitor controls leakage current. The nominal barrier height is a function of the difference between the work function of the electrode and the electron affinity of the dielectric. The electron affinity of a dielectric is closely related to the conduction band offset of the dielectric. The Schottky emission behavior of a dielectric layer is generally determined by the properties of the dielectric/electrode interface. Frenkel-Poole emission allows the conduction of charges through a dielectric layer through the interaction with defect sites such as vacancies, grain boundaries, and the like. As such, the Frenkel-Poole emission behavior of a dielectric layer is generally determined by the dielectric layer's bulk properties. Fowler-Nordheim emission allows the conduction of charges through a dielectric layer through direct tunneling without any intermediary interaction with e.g. defects. As such, the Fowler-Nordheim emission behavior of a dielectric layer is generally determined by the physical thickness of the dielectric layer. This leakage current is a primary driving force in the adoption of high-k dielectric materials. The use of high-k materials allows the physical thickness of the dielectric layer to be as thick as possible while maintaining the required capacitance (see Eqn 1 above).
High k metal oxide materials generally possess a number of defects in the form of oxygen vacancies. As an example, TiO2 is well known to exhibit n-type material properties due to the presence of oxygen vacancies. The defects contribute trap states within the band gap of the material and increase the leakage current due to the Frenkel-Poole emission mechanism discussed previously. Also, the carrier density is increased since the oxygen vacancies also act as donors. Annealing the dielectric material in an oxidizing atmosphere after deposition will improve the leakage current; however, oxygen vacancies will still be present. One method for further decreasing the leakage current in high k dielectric materials is to add dopants to the material to compensate the carriers generated from the defects or vacancies. Examples of dopants for TiO2 dielectric materials comprise Al, Ce, Co, Er, Ga, Gd, Ge, Hf, In, La, Lu, Mg, Mn, Nd, Pr, Sc, Si, Sn, Sr, Y, Zr, or combinations thereof. As an example, since TiO2 is an n-type material, acceptor dopants such as Al and Ge are often used to decrease the leakage current. These dopants are trivalent and neutralize the free carriers generated by the oxygen vacancies. They are believed to be substitutional donors for Ti when added to the TiO2. However, Al and Ge are small atoms, have limited solubility in TiO2, and their incorporation distorts the TiO2 matrix. Even after doping the TiO2 with Al or Ge, at elevated temperatures, free carriers again become available and the leakage current increases.
A second (or more) dopant has sometimes been added to the doped TiO2 to further decrease the leakage current. Typically, the second dopant has an ionic radius that is larger than the first dopant. However, the control of the composition and the control of the uniformity of the composition through the depth of the material are difficult using traditional processing schemes. Therefore, there is a need to develop methods for improving the incorporation and control of the dopants when two or more dopants are incorporated into a high k dielectric material.